The Detroit police chief admitted Monday that the facial recognition technology used by the department mistakenly identifies suspects about 96 percent of the time. It is a revealing admission as the Detroit Police Department faces criticism for arresting a man based on a bogus match in facial recognition software.
Last week, the ACLU filed a complaint with the Detroit Police Department on behalf of Robert Williams, a black man who was wrongfully arrested for stealing five watches worth $ 3,800 from a luxury retail store. The researchers first identified Williams by doing a facial recognition search with software from a company called DataWorks Plus. In police questioning, Williams noted that the grainy surveillance footage obtained by the police did not actually look like him. Police lacked other evidence linking Williams to the crime, so they reluctantly let him go.
Now Vice’s Jason Koebler reports that Detroit Police Chief James Craig recognized the flaws with his facial recognition software at a Monday event.
“If we only used the software [to identify subjects]We wouldn’t solve the case 95-97 percent of the time, “Craig said.” That is if we totally trust the software, which would be against our current policy … If we just used the technology on its own, to identify someone I would say 96 percent of the time they would misidentify themselves. “
As Craig points out, Detroit police officers are not supposed to arrest someone based solely on the results of a facial recognition search. And Detroit police said they didn’t do that in the Williams case. A police spokeswoman told The New York Times that “the investigator reviewed the video, interviewed witnesses, [and] made a lineup of photos “before arresting Williams.
But this “investigation” was quite flimsy. The “photo alignment” consisted of showing photos to a security contractor who was not an eyewitness; I’d only seen the same surveillance footage that the police had used in the first place.
Research has found that the accuracy of facial recognition software varies by subject’s race, and black suspects are less frequently identified than whites. And Koebler points out that DPD’s own statistics show that the technology is used almost exclusively on black suspects. According to police data, 68 out of 70 facial recognition searches were conducted on black suspects, while two had a race code of “U”, probably short for “unknown.”
The ACLU has called on the Detroit Police Department – and other police departments – to stop using facial recognition technology for investigations in light of its high error rate and racially disparate impact. The Boston city council voted to ban the use of facial recognition technology last week.